#### R255 Theory of Deep Learning # Module Introduction <!-- Put the link to this slide here so people can follow --> Ferenc Huszár (fh277) and Challenger Mishra (cm2099) --- ## Reading Group Format * **key content:** reading list (Moodle) * please read papers * refresh relevant background knowledge * take active part in discussions --- ## How to Read a Paper * [Keshav](http://ccr.sigcomm.org/online/files/p83-keshavA.pdf): How to read a paper * three-step process * **pass 1:** titles, intro, conclusions * **pass 2:** figures+captions, background * **pass 3:** the details --- ## Background Knowledge * you will need it to understand paper * Wikipedia often suffices * prioritize: skip proofs or details selectively. * ask others/us for background reading if unsure --- ## Presentations * two per session * talk + Q&A ≈ 25 minutes * aim for 15 mins for talk * speakers are randomly assigned * limited trades possible --- ## What is the Talk About * background! * motivating this work * key concepts or novelty * findings * remaining open questions * personal opinion/criticism * optional: follow-up work since then * skip proofs and details **unless useful** ---  --- ### Coursework * group projects * 10 groups of 3 * each group works on different topic * each person has specific sub-project * mini-viva * **deadline: 17 Jan 2023** * intermediate deadline: Week 5 --- ## Q and A Ferenc: fh277@cam.ac.uk Challenger: cm2009@cam.ac.uk